Conclusion
People don’t give much thought to what’s inside a charger,
but a lot of interesting circuitry is crammed inside.
The charger uses advanced techniques such as power factor correction and a resonant switching power supply to produce 85 watts of power in a compact, efficient unit.
The Macbook charger is an impressive piece of engineering, even if it’s not as reliable as you’d hope.
On the other hand, cheap no-name chargers cut corners and often have safety issues, making them risky, both to you and your computer.
Notes and references
The main alternative to a switching power supply is a linear power supply, which is much simpler and converts excess voltage to heat.
Because of this wasted energy, linear power supplies are only about 60% efficient, compared to about 85% for a switching power supply.
Linear power supplies also use a bulky transformer that may weigh multiple pounds, while switching power supplies can use a tiny high-frequency transformer.
Switching power supplies were taking over the computer industry as early as 1971. Electronics World said that
companies using switching regulators “read like a ‘Who’s Who’ of the computer industry: IBM, Honeywell, Univac, DEC, Burroughs, and RCA, to name a few”. See
“The Switching Regulator Power Supply”, Electronics World v86 October 1971, p43-47.
In 1976, Silicon General introduced SG1524 PWM integrated circuit, which put the control circuitry for a switching power supply on a single chip.
The quote about the Apple II power supply is from page 74 of the 2011 book
Steve Jobs by Walter Isaacson.
Steve Job’s quote sounds convincing, but I consider it the reality distortion field in effect.
If anyone can take the credit for making switching power supplies an inexpensive everyday product, it is Robert Boschert.
He started selling switching power supplies in 1974
for everything from printers and computers to the F-14 fighter plane.
See Robert Boschert: A Man Of Many Hats Changes The World Of Power Supplies in Electronic Design.
The Apple II’s power supply is very similar to the
Boschert OL25 flyback power supply but with a patented variation.
You might expect the bad power factor is because switching power supplies rapidly turn on and off, but that’s not the problem. The difficulty comes from the nonlinear diode bridge, which charges the input capacitor only at peaks of the AC signal.
(If you’re familiar with power factors due to phase shift, this is totally different. The problem is the non-sinusoidal current, not a phase shift.)
The idea behind PFC is to use a DC-DC boost converter before the switching power supply itself. The boost converter is carefully controlled so its input current is a sinusoid proportional to the AC waveform. The result is the boost converter looks like a nice resistive load to the power line, and the boost converter supplies steady voltage to the switching power supply components.
The charger uses a MC33368 “High Voltage GreenLine Power Factor Controller” chip
to run the PFC. The chip is designed for low power, high-density applications so it’s a good match for the charger.
The SMPS controller chip is a L6599 high-voltage resonant controller; for some reason it is labeled DAP015D.
It uses a resonant half-bridge topology; in a half-bridge circuit, two transistors control power through the transformer first one direction and then the other.
Common switching power supplies use a PWM (pulse width modulation) controller, which adjusts the time the input is on.
The L6599, on the other hand, adjusts the frequency instead of the pulse width. The two transistors alternate switching on for 50% of the time. As the frequency increases above the resonant frequency, the power drops, so controlling the frequency regulates the output voltage.
The processor in the charger is a MSP430F2003 ultra low power microcontroller with 1kB of flash and just 128 bytes of RAM. It includes a high-precision 16-bit analog to digital converter. More information is here.
The 68000 microprocessor from the original Apple Macintosh and the 430 microcontroller in the charger aren’t directly comparable as they have very different designs and instruction sets.
But for a rough comparison, the 68000 is a 16/32 bit processor running at 7.8MHz, while the MSP430 is a 16 bit processor running at 16MHz.
The Dhrystone benchmark measures
1.4 MIPS (million instructions per second) for the 68000 and much higher performance of
4.6 MIPS for the MSP430. The MSP430 is designed for low power consumption, using about 1% of the power of the 68000.
The 60W Macbook charger uses a custom MSP430 processor, but the 85W charger uses a general-purpose processor that needs to loaded with firmware.
The chip is programmed with the Spy-Bi-Wire interface, which is TI’s two-wire variant of the standard JTAG interface.
After programming, a security fuse inside the chip is blown to prevent anyone from reading or modifying the firmware.
The voltage to the processor is provided by not by a standard voltage regulator, but a LT1460 precision reference, which outputs 3.3 volts with the exceptionally high accuracy of 0.075%. This seems like overkill to me;
this chip is the second-most expensive chip in the charger after the SMPS controller, based on Octopart’s prices.
The voltage reference chip is unusual, it is a TSM103/A that combines two op amps and a 2.5V reference in a single chip.
Semiconductor properties vary widely with temperature, so keeping the voltage stable isn’t straightforward.
A clever circuit called a bandgap reference cancels out temperature variations;
Since some readers are very interested in grounding, I’ll give more details.
A 1KΩ ground resistor connects the AC ground pin to the charger’s output ground. (With the 2-pin plug, the AC ground pin is not connected.)
Four 9.1MΩ resistors connect the internal DC ground to the output ground. Since they cross the isolation boundary, safety is an issue. Their high resistance avoids a shock hazard. In addition, since there are four resistors in series for redundancy, the charger remains safe even if a resistor shorts out somehow. There is also a Y capacitor (680pF, 250V) between the internal ground and output ground; this blue capacitor is on the upper side of the board. A T5A fuse (5 amps) protects the output ground.
The power in watts is simply the volts multiplied by the amps. Increasing the voltage is beneficial because it allows higher wattage; the maximum current is limited by the wire size.
The control circuitry is fairly complex.
The output voltage is monitored by an op amp in the TSM103/A chip which compares it with a reference voltage generated by the same chip. This amplifier sends a feedback signal via an optoisolator to the SMPS control chip on the primary side. If the voltage is too high, the feedback signal lowers the voltage and vice versa. That part is normal for a power supply, but ramping the voltage from 16.5 volts to 18.5 volts is where things get complicated.
The output current creates a voltage across the current sense resistors, which have a tiny resistance of 0.005Ω each – they are more like wires than resistors. An op amp in the TSM103/A chip amplifies this voltage. This signal goes to tiny TS321 op amp which starts ramping up when the signal corresponds to 4.1A. This signal goes into the previously-described monitoring circuit, increasing the output voltage.
The current signal also goes into a tiny TS391 comparator, which sends a signal to the primary through another optoisolator to cut the output voltage. This appears to be a protection circuit if the current gets too high.
The circuit board has a few spots where zero-ohm resistors (i.e. jumpers) can be installed to change the op amp’s amplification. This allows the amplification to be adjusted for accuracy during manufacture.
If you measure the voltage from a Macbook charger, you’ll find about six volts instead of the 16.5 volts you’d expect. The reason is the output is deactivated and you’re only measuring the voltage through the bypass resistor just below the output transistor.
The laptop pulls the charger output low with a 39.41KΩ resistor to indicate that it is ready for power. An interesting thing is it won’t work to pull the output too low – shorting the output to ground doesn’t work. This provides a safety feature. Accidental contact with the pins is unlikely to pull the output to the right level, so the charger is unlikely to energize except when properly connected.
The imitation charger uses the Fairchild FAN7602 Green PWM Controller chip, which is more advanced than I expected in a knock-off;
I wouldn’t have been surprised if it just used a simple transistor oscillator.
Another thing to note is the imitation charger uses a single-sided circuit board, while the genuine uses a double-sided circuit board, due to the much more complex circuit.
The burnt charger is an Apple A1222 85W Macbook charger, which is a different model from the A1172 charger in the rest of the teardown.
The A1222 is in a slightly smaller, square case and has a totally different design based on the
NCP 1203 PWM controller chip.
Components in the A1222 charger are packed even more tightly than in the A1172 charger. Based on the burnt-up charger, I think they pushed the density a bit too far.
[19]
I looked up many of the charger components on Octopart to see their prices. Apple’s prices should be considerably lower. The charger has many tiny resistors, capacitors and transistors; they cost less than a cent each. The larger power semiconductors, capacitors and inductors cost considerably more. I was surprised that the 16-bit MSP430 processor costs only about $0.45. I estimated the price of the custom transformers. The list below shows the main components.
| Component | Cost | |
|---|---|---|
| MSP430F2003 processor | $0.45 | |
| MC33368D PFC chip | $0.50 | |
| L6599 controller chip | $1.62 | |
| LT1460 3.3V reference | $1.46 | |
| TSM103/A reference | $0.16 | |
| 2x P11NM60AFP 11A 600V MOSFET | $2.00 | |
| 3x Vishay optocoupler | $0.48 | |
| 2x 630V 0.47uF film capacitor | $0.88 | |
| 4x 25V 680uF electrolytic capacitor | $0.12 | |
| 420V 82uF electrolytic capacitor | $0.93 | |
| polypropylene X2 capacitor | $0.17 | |
| 3x toroidal inductor | $0.75 | |
| 4A 600V diode bridge | $0.40 | |
| 2x dual common-cathode schottky rectifier 60V, 15A | $0.80 | |
| 20NC603 power MOSFET | $1.57 | |
| transformer | $1.50? | |
| PFC inductor | $1.50? |
The article Breaking down the full $650 cost of the iPhone 5 describes Apple’s profit margins in detail, estimating 45% profit margin on the iPhone.
Some people have suggested that Apple’s research and development expenses explain the high cost of their chargers, but the math shows R&D costs must be negligible.
The book Practical Switching Power Supply Design estimates 9 worker-months to design and perfect a switching power supply, so perhaps $200,000 of engineering cost. More than 20 million Macbooks are sold per year, so the R&D cost per charger would be one cent. Even assuming the Macbook charger requires ten times the development of a standard power supply only increases the cost to 10 cents.
Originally Authored by Ken Shirriff at his Blog.
[tps_footer][/tps_footer]
